wf-recorder - Man Page

wf-recorder is a tool built to record your screen on Wayland compositors. It makes use of wlr-screencopy for capturing video and ffmpeg(1) for encoding it.

In its simplest form, run wf-recorder to start recording and use ‘Ctrl+C’ to stop. This will create a file called ‘recording.mkv’ in the current working directory using the default codec.

The options are as follows:

-a, --audio [=DEVICE]

Starts recording the screen with audio.

DEVICE argument is optional. In case you want to specify the PulseAudio device which will capture the audio, you can run this command with the name of that device. You can find your device by running

$ pactl list sources | grep Name

-b, --bframes max_b_frames

Sets the maximum number of B-Frames to use.

-B, --buffrate buffrate

Tells the encoder a prediction of what framerate to expect. This preserves VFR and Solves FPS limit issue of some encoders (like svt-av1). Should be set to the same framerate as display.

-c, --codec output_codec

Specifies the codec of the video. Supports GIF output as well.

To modify codec parameters, use -p option_name=option_value

-r, --framerate framerate

Sets hard constant framerate. Will duplicate frames to reach it. This makes the resulting video CFR. Solves FPS limit issue of some encoders.

-d, --device encoding_device

Selects the device to use when encoding the video.

Some drivers report support for ‘rgb0’ data for vaapi input but really only support yuv. Use the -x yuv420 option in addition to the vaapi options to convert the data in software, before sending it to the GPU.

--no-dmabuf

By default, wf-recorder will try to use only GPU buffers and copies if using a GPU encoder. However, this can cause issues on some systems. In such cases, this option will disable the GPU copy and force a CPU one.

-D, --no-damage

By default, wf-recorder will request a new frame from the compositor only when the screen updates. This results in a much smaller output file, which however has a variable refresh rate. When this option is on, wf-recorder does not use this optimization and continuously records new frames, even if there are no updates on the screen.

-f filename.ext

By using the -f option, the output file will have the name filename.ext and the file format will be determined by the provided extension. If the extension is not recognized by your ffmpeg(1) muxers, the command will fail.

You can check the muxers that your ffmpeg(1) installation supports by running

$ ffmpeg -muxers

-F, --filter filter_string

Set the ffmpeg filter to use. VAAPI requires `scale_vaapi=format=nv12:out_range=full` to work.

-g, --geometry screen_geometry

Selects a specific part of the screen. The format is "x,y WxH".

-h, --help

Prints the help screen.

-l, --log

Generates a log on the current terminal. For debug purposes.

-m, --muxer muxer

Set the output format to a specific muxer instead of detecting it from the filename.

-o, --output

Specify the output where the video is to be recorded.

-p, --codec-param [option_name=option_value]

Change the codec parameters.

-v, --version

Print the version of wf-recorder.

-x, --pixel-format pixel_format

Set the output pixel format.

List available formats using

$ ffmpeg -pix_fmts

--audio-backend audio_backend

Specifies the audio backend to be used when -a is set.

-C, --audio-codec output_audio_codec

Specifies the codec of the audio.

-P, --audio-codec-param [option_name=option_value]

Change the audio codec parameters.

-R, --sample-rate sample_rate

Changes the audio sample rate, in HZ. The default value is 48000.

-X, --sample-format sample_format

Set the output audio sample format.

List available formats using

$ ffmpeg -sample_fmts

-y, --overwrite

Force overwriting the output file without prompting.

To select a specific part of the screen you can either use --g geometry or use https://github.com/emersion/slurp for interactive selection of the screen area that will be recorded:

$ wf-recorder -g $(slurp)

You can record screen and sound simultaneously with

$ wf-recorder --audio --file=recording_with_audio.mkv

To specify an audio device, use the --a<DEVICE> or ---audio=<DEVICE> options.

To specify a codec use the -c codec option. To modify codec parameters, -p option_name=option_value.

To set a specific output format, use the -m, --muxer option. For example, to output to a video4linux2 loopback you might use:

$ wf-recorder --muxer=v4l2 --codec=rawvideo --file=/dev/video2

To use GPU encoding, use a VAAPI codec (for ex. ‘vp9_vaapi’ ) and specify a GPU device to use with the -d option:

$ wf-recorder -f test-vaapi.mkv -c vp9_vaapi -d /dev/dri/renderD128

Some drivers report support for ‘rgb0’ data for ‘vaapi’ input but really only support yuv planar formats. In this case, use the -x yuv420p option in addition to the ‘vaapi’ options to convert the data to yuv planar data before sending it to the GPU.

ffmpeg(1), pactl(1)